[unreadable] We propose to extend our successful development of an agile spectral light source for light microscopy, funded through the NCI-IMAT initiative, into FDA trials. The SpectraLamp(tm) device enables the automated and quantitative analysis of double-immunostained samples in brightfield (non-fluorescence-based) microscopy, with particular utility for hematopathology applications. A clinically compelling area is the enumeration of malignant blasts in bone marrow biopsy specimens of patients with acute leukemias, myelodysplastic syndromes and chronic myleloproliferative diseases for the purpose of staging and evaluating therapeutic responses. Current methods are inadequate due to poor sampling (typical of bone marrow aspirates) or difficulty in identifying true blasts (bone marrow biopsies and single-color immunohistochemical phenotyping). Double-immunophenotyping permits largely unambiguous detection, but such labeling strategies are not supported by standard (non-multispectral) color imaging systems. To overcome this limitation, we will combine our multispectral imaging system with a high-speed slide-scanning platform that can scan an entire bone-marrow biopsy at high resolution in under 3 minutes. Machine-learning software tools and spectral imaging will identify blasts, combining morphology parameters and double or even triple immunophenotyping to ensure accuracy and precision. After blasts are identified, flow-cytometrylike software will present the data in histograms and other modalities, allowing the user to "gate" on particular cellular populations and pull up panels of cell images for confirmation. [unreadable] [unreadable] Time-line: Final equipment-related tasks will be accomplished in the first year and preliminary testing of the instrumentation and software features will occur in the second year, as will development of qualified panels of immunoreagents and staining protocols (by DAKOCytomation). FDA-sanctioned clinical trials will be conducted in the third year, leading to submission of an application for a 510(k) clearance. [unreadable] [unreadable]